54 (2) • May 2005: 411–425 Oxelman & al. • Disintegration of Scrophulariaceae Further disintegration of Scrophulariaceae Bengt Oxelman1, Per Kornhall1, Richard G. Olmstead2 & Birgitta Bremer3 1 Department of Systematic Botany, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18D SE- 752 36 Uppsala, Sweden. [email protected]. (author for correspondence), [email protected] 2 Department of Biology, Box 355325, University of Washington, Seattle, Washington 98195, U.S.A. olm- [email protected] 3 The Bergius Foundation at the Royal Swedish Academy of Sciences, Box 50017 SE-104 05 Stockholm, Sweden. [email protected] A phylogenetic study of plastid DNA sequences (ndhF, trnL/F, and rps16) in Lamiales is presented. In partic- ular, the inclusiveness of Scrophulariaceae sensu APG II is elaborated. Scrophulariaceae in this sense are main- ly a southern hemisphere group, which includes Hemimerideae (including Alonsoa, with a few South American species), Myoporeae, the Central American Leucophylleae (including Capraria), Androya, Aptosimeae, Buddlejeae, Teedieae (including Oftia, Dermatobotrys, and Freylinia), Manuleeae, and chiefly Northern tem- perate Scrophularieae (including Verbascum and Oreosolen). Camptoloma and Phygelius group with Buddlejeae and Teedieae, but without being well resolved to any of these two groups. Antherothamnus is strongly supported as sister taxon to Scrophularieae. African Stilbaceae are shown to include Bowkerieae and Charadrophila. There is moderate support for a clade of putative Asian origin and including Phrymaceae, Paulownia, Rehmannia, Mazus, Lancea, and chiefly parasitic Orobanchaceae, to which Brandisia is shown to belong. A novel, strongly supported, clade of taxa earlier assigned to Scrophulariaceae was found. The clade includes Stemodiopsis, Torenia, Micranthemum and probably Picria and has unclear relationships to the rest of Lamiales. This clade possibly represents the tribe Lindernieae, diagnosed by geniculate anterior filaments, usually with a basal swelling. KEYWORDS: Lamiales, ndhF, phylogeny, rps16, Scrophulariaceae, trnL/F. influential classifications for the 19th century concept of INTRODUCTION Scrophulariaceae were those of Bentham (1846), who in Lamiales sensu APG II (The Angiosperm Phylogeny addition recognized Selaginaceae Choisy, Wettstein Group, 2003) constitute a major clade of flowering (1895), who recognized Lentibulariaceae, Plantagin- plants, with approximately 17,800 species (Judd & al., aceae, and Orobanchaceae as separate families, and 2002) and 23 families, representing 12.3% of eudicot Hallier (1903), who made the broadest circumscription diversity (Stevens, 2001, onwards). They belong to a of the family. major asterid clade, the Lamiids (Bremer & al., 2002). Even if doubts sometimes were expressed regarding The monophyly of Lamiales is comparatively uncontro- the naturalness of Scrophulariaceae (e.g., Thieret, 1967; versial and well supported by molecular (e.g., Backlund Barringer, 1984), it was not until the study by Olmstead & al., 1998; Oxelman & al., 1999; Albach & al., 2001; & Reeves (1995) based on plastid ndhF and rbcL gene Bremer & al., 2002) and phytochemical data (Jensen, sequences that the concept of Scrophulariaceae needed a 1992; Scogin, 1992). Morphologically, they typically are revolutionary revision in order to fit within a phyloge- characterized by opposite leaves, sympetalous zygomor- netic framework. Olmstead & Reeves (1995) discovered phic flowers, oligosaccharides, frequent production of 6- two clearly separated clades consisting of scrophularia- oxygenated flavones, embryos of Onagrad type and cap- ceous taxa. One clade (“scroph I”) included Buddleja L., sular fruits. Although the majority of the many well- Selago L., Verbascum L. and the type genus known families in Lamiales are well supported (e.g., Scrophularia L., whereas the other (“scroph II”) includ- Olmstead & al., 2001), the relationships among them are ed Antirrhinum L., Digitalis L., Veronica L., Plantago L., obscure, despite several recent molecular studies (e.g., Hippuris L., and Callitriche L. In their study the enig- Olmstead & Reeves, 1995; Backlund & al., 1998; matic woody Paulownia Siebold & Zucc., which was Oxelman & al., 1999; Olmstead & al., 2001). usually classified in Scrophulariaceae but transferred to The concept “Scrophulariae” occurs in Durande, Bignoniaceae Juss. by Takthajan (1980), and Schlegelia Notions Elém. Bot.: 265. 1782, but the name of the fam- Miq., which was originally placed in Bignoniaceae but ily is conserved as Scrophulariaceae Jussieu. The most argued by Armstrong (1985) to fit better within 411 Oxelman & al. • Disintegration of Scrophulariaceae 54 (2) • May 2005: 411–425 Scrophulariaceae, were both left in uncertain positions accordance with phylogenetic relationships can be pre- within Lamiales. sented. Fischer (2004) has presented a tentative classifi- Several subsequent molecular phylogenetic studies cation of all genera belonging to Scrophulariaceae sensu focusing either on other taxa of Lamiales or more inclu- lato (i.e., approximately in the sense of Hallier, 1903) sive groups (e.g., Soltis & al., 1998; Olmstead & al., into tribes and informal higher groups (“families”). 2000, 2001; Albach & al., 2001; Bremer & al., 2002) The aims of this study are to identify the inclusive- have confirmed the general pattern revealed by Olmstead ness of the “scroph I” clade, sensu Olmstead & Reeves & Reeves (1995). (1995), using plastid DNA sequence data and to infer The parasitic plants that have variously been placed relationships of representatives of major groups of taxa in Scrophulariaceae or Orobanchaceae Vent. have been previously assigned to Scrophulariaceae, for which plas- shown to constitute an additional monophyletic group tid DNA sequences have not been obtained before. (dePamphilis & al., 1997; Wolfe & dePamphilis, 1998; Young & al., 1999). Olmstead & al. (2001) merged the rbcL/ndhF dataset and the data from the plastid rps2 gene that had been useful for the recognition of MATERIAL AND METHODS Orobanchaceae. They also added some other taxa, and Sampling. — Most previous molecular studies were thereby able to identify an additional distinct scro- have used sequences of rbcL and ndhF, but many other phulariaceous clade, Calceolariaceae. In addition, plastid loci have been introduced as well, including rps2 Halleria L. was shown to group with Stilbaceae Kunth, (dePamphilis & al., 1997), trnL/F intron/spacer region and Mimulus L. did not belong to any of these five clades (Freeman & Scogin, 1999; McDade & Moody, 1999), of ex-Scrophulariaceae taxa. Beardsley & Olmstead rps16 (Wallander & Albert, 2000; Bremer & al., 2002), (2002) identified Mimulus as a member of another clade, matK (Bremer & al., 2002), trnT/F spacer region, and the Phrymaceae. Globularia L. has been demonstrated to trnV spacer (Bremer & al., 2002). We use the ndhF, belong to the Veronicaceae clade sensu Oxelman & al. trnL/F, and rps16 DNA sequence regions, partly because (1999), Olmstead & al. (2001), Kornhall & al. (2001), a substantial number of bulk Lamiales taxa have already and Bremer & al. (2002). been sequenced for these, and partly because previous Olmstead & al. (2001) recognized five distinct phy- studies have indicated that these regions are particularly logenetic lineages composed mainly of taxa previously informative in Lamiales. Initially, we included all assigned to Scrophulariaceae. Nevertheless, several of Lamiales taxa with sequence information from either of their included genera (e.g., Schlegelia, Paulownia, the trnL/F, rps16, or ndhF regions deposited in EMBL/ Mimulus) did not group with any of these five lineages. Genbank as per May 2002. However, because of uneven This fact, together with the restricted sampling (39 of ca. representation, we used an exemplar approach for groups 280 genera in Scrophulariaceae in a traditional sense, whose monophyly is not in question. In addition, we (see Watson & Dallwitz, 1992 onwards), calls for more tried to obtain material for DNA extraction from repre- extensive sampling of genera previously assigned to sentatives of genera that have been suggested to be relat- Scrophulariaceae. ed to Scrophulariaceae sensu APG II. We also included Scrophulariaceae sensu APG II (2003) constitutes representatives from other parts of Scrophulariaceae approximately what Olmstead & Reeves (1995) identi- sensu Hallier and from Stilbaceae. We tried to find addi- fied as the “scroph I” clade, that is, Buddlejaceae K. tional representatives of every taxon, and sequence at Wilh. (see Oxelman & al., 1999 for phylogenetic cir- least one of the three regions to verify the taxonomic cumscription), Manuleeae Benth. (incl. Selagineae, see identity of the sequences. Information on sampled taxa, Kornhall & al., 2001), the two large genera Scrophularia and EMBL/Genbank accession numbers can be found in and Verbascum, plus the tribe Hemimerideae Benth. and the Appendix. Myoporaceae R. Br., including Androya Perrier (Oxel- Sequencing. — Most of the sequencing was per- man & al., 1999; Bremer & al., 2002). The assignment of formed either at the Evolutionary Biology Centre, Hemimerideae to this clade has received only weak sup- Uppsala University, Sweden, or at the Department of port in previous studies. Several taxa that have been sug- Botany, University of Washington, Seattle, U.S.A. Total gested to have affinities with Myoporaceae (e.g., DNA was usually extracted from dried plant material Capraria L., Anticharis Endl.